In einer eMail vom 11.03.00 09:20:12 (MEZ) Mitteleuropäische Zeit schreibt 
[EMAIL PROTECTED]:

> Hi Volker,
>  
>  This is kind of  off target here, but..
>  The method below would solve that problem in rebol/view. It ensures that
>  processes don't keep running in the task manager after the window is closed
>  for a rebol/view script that has an infinite loop.
>  Hopefully the event scheme could be added to rebol/core too. or Perhaps 
some
>  of the rebol/core gurus have similar solution for rebol/core?
>  
>  event-port: open [scheme: 'event]
>  forever [
>      ; do stuff...
>      ; do more stuff
>      wait event-port ; since time events are always firing it won't have to
>  wait long.
>      event: first event-port
>      if event/type = 'quit [break]
>      do event
>  ]
>  close event-port
>  ; end of script..
>  

Hm, this is not asynchrone?
i need a "programmable escape key". The problem is, what i do
with _buggy_ infinite loops when i test over the net (the "do stuff" hangs, 
not the main-loop)? 
there is no escape-key
then. Think i want _really_ to use one of this nice rebol-cgi-servers, or, 
Yea,
the wicki server (:-)), and my script has a bug. 
happily the server hangs.
if i could say [after 5 seconds escape with 'server-restart], it would be 
much safer.
running the server in a extern shell-loop, it could free itself nearly 
allways.
having memory-limits, context-protection (Gabrieles?) too and a really 
crash-hard kernel 
would make this stuff really portable, but a simple "start: rebol server.r ^/ 
goto start" 
is so short that porting is acceptable :)
Killing rebols, especially on windows, is another thing. for example, in my 
perl
"kill" does not work, and external communication would be needed to say 
"ok, iam well, let me continue".
Having a "escape-key" and controlled pre-buffering would make more threading 
nearly
unneccessary, i think. (Maybe full threading looks cool, but, write Java with 
and without it..
And than look at Oberon (well, partly:). And think you write REBOL in 
minutes, not java-like system-design :)
Hopefully iam missing something :)
Note: the realtime JBed (mentioned in Dr Dobbs) preferres this approach..
Note2: the original message was "can't stop rebol, bad script"..

>  Cheers,
>  
>  Allen K

Cheers back :)

Volker

Reply via email to